A library to migrate Elasticsearch mappings. Inspired by flyway.

Elasticsearch-Evolution executes versioned migration scripts reliable and persists the execution state in an internal Elasticsearch/Opensearch index. Successful executed migration scripts will not be executed again!

• tested on Java 17 and 21
• runs on Spring-Boot 3.x (and of course without Spring-Boot)
• runs on Elasticsearch version 7.5.x - 8.13.x
• runs on Opensearch version 1.x and 2.x
• highly configurable (e.g. location(s) of your migration files, migration files format pattern)
• placeholder substitution in migration scripts
• easily extendable to your needs
• supports microservices / multiple parallel running instances via logical database locks
• ready to use default configuration
• line comments in migration files

First add the latest version of Elasticsearch-Evolution spring boot starter as a dependency in your maven pom.xml:

<dependency>
    <groupId>com.senacor.elasticsearch.evolution</groupId>
    <artifactId>spring-boot-starter-elasticsearch-evolution</artifactId>
    <version>0.6.0</version>
</dependency>

Elasticsearch-Evolution uses internally Elastics RestClient and requires at minimum version 7.5.2. Spring boot could use an older version, depending on your Spring Boot version, so update it in your pom.xml:

<properties>
    <elasticsearch.version>7.5.2</elasticsearch.version>
</properties>

Place your migration scripts in your application classpath at es/evolution

That's it. Elasticsearch-Evolution runs at application startup and expects your Elasticsearch/Opensearch at http://localhost:9200

First add the latest version of Elasticsearch-Evolution core as a dependency:

<dependency>
    <groupId>com.senacor.elasticsearch.evolution</groupId>
    <artifactId>elasticsearch-evolution-core</artifactId>
    <version>0.6.0</version>
</dependency>

Place your migration scripts in your application classpath at es/evolution

Create a ElasticsearchEvolution instance and execute the migration.

// first create a Elastic RestClient
RestClient restClient = RestClient.builder(HttpHost.create("http://localhost:9200")).build();
// then create a ElasticsearchEvolution configuration and create a instance of ElasticsearchEvolution with that configuration
ElasticsearchEvolution elasticsearchEvolution = ElasticsearchEvolution.configure()
        .load(restClient);
// execute the migration
elasticsearchEvolution.migrate();

A Elasticsearch-Evolutions migration script represents just a rest call. Here is an Example:

PUT _template/my_template
Content-Type: application/json

{
  "index_patterns": [
    "my_index_*"
  ],
  "order": 1,
  "version": 1,
  "settings": {
    "number_of_shards": 1
  },
  "mappings": {
    "properties": {
    "version": {
      "type": "keyword",
      "ignore_above": 20,
      "similarity": "boolean"
    },
    "locked": {
      "type": "boolean"
    }
    }
  }
}

The first line defines the HTTP method PUT and the relative path to the Elasticsearch/Opensearch endpoint _template/my_template to create a new mapping template. Followed by a HTTP Header Content-Type: application/json. After a blank line the HTTP body is defined.

The pattern is strongly oriented in ordinary HTTP requests and consist of 4 parts:

1. The HTTP method (required). Supported HTTP methods are GET, HEAD, POST, PUT, DELETE, OPTIONS and PATCH. The First non-comment line must always start with a HTTP method.
2. The path to the Elasticsearch/Opensearch endpoint to call (required). The path is separated by a blank from the HTTP method. You can provide any query parameters like in a ordinary browser like this /my_index_1/_doc/1?refresh=true&op_type=create
3. HTTP Header(s) (optional). All non-comment lines after the HTTP method line will be interpreted as HTTP headers. Header name and content are separated by :.
4. HTTP Body (optional). The HTTP Body is separated by a blank line and can contain any content you want to sent to Elasticsearch/Opensearch.

Elasticsearch-Evolution supports line-comments in its migration scripts. Every line starting with # or // will be interpreted as a comment-line. Comment-lines are not send to Elasticsearch/Opensearch, they will be filtered by Elasticsearch-Evolution.

Elasticsearch-Evolution supports named placeholder substitution. Placeholders are marked in your migration script like this: ${my-placeholder}

• starts with placeholderPrefix which is by default ${ and is configurable.
• followed by the placeholder name which can be any string, but must not contain placeholderPrefix or placeholderSuffix
• ends with placeholderSuffix which is by default } and is configurable.

Here is an example filename: V1.0__my-description.http

The filename has to follow a pattern:

• starts with esMigrationPrefix which is by default V and is configurable.
• followed by a version, which have to be numeric and can be structured by separating the version parts with .
• followed by the versionDescriptionSeparator: __
• followed by a description which can be any text your filesystem supports
• ended with esMigrationSuffixes which is by default .http and is configurable and case-insensitive.

Elasticsearch-Evolution uses the version for ordering your scripts and enforces strict ordered execution of your scripts, by default. Out-of-Order execution is supported, but disabled by default. Elasticsearch-Evolution interprets the version parts as Integers, so each version part must be between 1 (inclusive) and 2,147,483,647 (inclusive).

Here is an example which indicates the ordering: 1.0.1 < 1.1 < 1.2.1 < (2.0.0 == 2). In this example version 1.0.1 is the smallest version and is executed first, after that version 1.1, 1.2.1 and in the end 2. 2 is the same as 2.0 or 2.0.0 - so trailing zeros will be trimmed.

NOTE: Versions with major version 0 are reserved for internal usage, so the smallest version you can define is 1

Elasticsearch-Evolution can be configured to your needs:

• enabled (default=true): Whether to enable or disable Elasticsearch-Evolution.
• locations (default=[classpath:es/migration]): List of locations of migrations scripts. Supported is classpath:some/path and file:/some/path. The location is scanned recursive, but only to a depth of 10. NOTE: all scripts in all locations / subdirectories will be flatted and only the version number will be used to order them.
• encoding (default=UTF-8): Encoding of migration files.
• defaultContentType (default=application/json; charset=UTF-8): This content type will be used as default if no contentType header is specified in the header section of a migration script. If no charset is defined, the encoding charset is used.
• esMigrationPrefix (default=V): File name prefix for migration files.
• esMigrationSuffixes (default=[.http]): List of file name suffixes for migration files. The suffix is checked case-insensitive.
• placeholderReplacement (default=true): Whether to enable or disable placeholder replacement in migration scripts.
• placeholders (default=[]): Map of placeholders and their replacements to apply to migration scripts.
• placeholderPrefix (default=${): Prefix of placeholders in migration scripts.
• placeholderSuffix (default=}): Suffix of placeholders in migration scripts.
• historyIndex (default=es_evolution): Name of the history index that will be used by Elasticsearch-Evolution. In this index Elasticsearch-Evolution will persist his internal state and tracks which migration script has already been executed.
• historyMaxQuerySize (default=1000): The maximum query size while validating already executed scripts. This query size have to be higher than the total count of your migration scripts.
• validateOnMigrate (default=true): Whether to fail when a previously applied migration script has been modified after it was applied.
• baselineVersion (default=1.0): Version to use as a baseline. versions lower than it will not be applied.
• lineSeparator (default=\n): Line separator, used only temporary between reading raw migration file line-by-line and parsing it later. Only needed for backward compatibility / checksum stability! Should be one of \n, \r or \r\n
• outOfOrder (default=false): Allows migrations to be run "out of order". If you already have versions 1.0 and 3.0 applied, and now a version 2.0 is found, it will be applied too instead of being rejected.
• trimTrailingNewlineInMigrations (default=false): Whether to remove a trailing newline in migration scripts. Only needed for backward compatibility / checksum stability!

You can set the above configurations via Spring Boots default configuration way. Just use the prefix spring.elasticsearch.evolution. Here is an example application.properties:

spring.elasticsearch.evolution.locations[0]=classpath:es/migration
spring.elasticsearch.evolution.locations[1]=classpath:es/more_migration_scripts
spring.elasticsearch.evolution.placeholderReplacement=true
spring.elasticsearch.evolution.placeholders.indexname=myIndexReplacement
spring.elasticsearch.evolution.placeholders.docType=_doc
spring.elasticsearch.evolution.placeholders.foo=bar
spring.elasticsearch.evolution.historyIndex=es_evolution

Since spring boot 2.1 AutoConfiguration for Elasticsearchs REST client is provided (see org.springframework.boot.autoconfigure.elasticsearch.ElasticsearchRestClientAutoConfiguration). You can configure the RestClient, required for Elasticsearch-Evolution, just like that in your application.properties:

spring.elasticsearch.uris[0]=https://example.com:9200
spring.elasticsearch.username=my-user-name
spring.elasticsearch.password=my-secret-pw

Elasticsearch-Evolutions just needs a RestClient as spring bean. If you don't have spring boot 2.1 or later or you need a special RestClient configuration e.g. to accept self signed certificates or disable hostname validation, you can provide a custom RestClient like this:

@Bean
public RestClient restClient() {
    RestClientBuilder builder = RestClient.builder(HttpHost.create("https://localhost:9200"))
            .setHttpClientConfigCallback(httpClientBuilder -> {
                        CredentialsProvider credentialsProvider = new BasicCredentialsProvider();
                        credentialsProvider.setCredentials(AuthScope.ANY, new UsernamePasswordCredentials("my-user-name", "my-secret-pw"));
                        httpClientBuilder.setDefaultCredentialsProvider(credentialsProvider);
                        try {
                            httpClientBuilder
                                    .setSSLContext(new SSLContextBuilder().loadTrustMaterial(null, TrustAllStrategy.INSTANCE).build())
                                    .setSSLHostnameVerifier(NoopHostnameVerifier.INSTANCE);
                        } catch (GeneralSecurityException e) {
                            throw new IllegalStateException("could not configure http client to accept all certificates", e);
                        }
                        return httpClientBuilder;
                    }
            );
    return builder.build();
}

Maybe you want to provide a customised Initializer for Elasticsearch-Evolution e.g with another order:

@Bean
public ElasticsearchEvolutionInitializer customElasticsearchEvolutionInitializer(ElasticsearchEvolution elasticsearchEvolution) {
    return new ElasticsearchEvolutionInitializer(elasticsearchEvolution) {
        @Override
        public int getOrder() {
            return Ordered.LOWEST_PRECEDENCE;
        }
    };
}

You can set the above configurations via the ElasticsearchEvolutionConfig fluent builder like this:

ElasticsearchEvolution.configure()
    .setLocations(Collections.singletonList("classpath:es/migration"))
    .setPlaceholderReplacement(true)
    .setPlaceholders(Collections.singletonMap("indexname", "myIndexReplacement"))
    .setHistoryIndex("es_evolution");

• added regression tests against OpenSearch 2.14
• bump spring boot version to 3.3.0
• added regression tests for spring boot 3.3

• Added option to trim a trailing newline in migration scripts (fixes #298). NOTE: This option is only needed for backward compatibility / checksum stability!
• The minimum supported Java version is now 17.
• Drop spring boot 2 compatibility. Further versions may run on spring boot 2, but it is not tested anymore.

• bugfix (#293): trailing newlines will no longer be removed from migration scripts.
• added regression tests against OpenSearch 2.13
• added regression tests against ElasticSearch 8.13

• version updates (spring-boot 2.7.18)
• added regression tests for spring boot 3.2
• remove deprecated query parameter ignore_throttled from ES requests

• added spring boot configuration metadata #240
• replaces unmaintained org.reflections library with classgraph to scan the classpath for migration files. Fixes #239

• support out of order migration execution.
• version updates (spring-boot 2.7.17)
• added regression tests against OpenSearch 2.11, 2.10 and 2.9
• added regression tests against ElasticSearch 8.11. 8.10 and 8.9
• drop older Elasticsearch and OpenSearch versions in regression tests. Only test against the last 3 minor versions of the latest major release.
• added regression tests on JDK 21
• added regression tests for spring boot 3.1
• update org.reflections:reflections from 0.9.12 to 0.10.2 #233 thanks @RiVogel
• KNOWN ISSUES:
  • #239: Migration files not found in Spring Boot jar
    • Workaround 1: downgrade org.reflections:reflections to 0.10.1
    • Workaround 2: downgrade elasticsearch-evolution to 0.4.2

• bugfix (#182): checksum calculation was based on system dependent line separators which lead to different checksums on different operating systems (e.g. windows vs linux). The default is now \n. For backward compatibility you can set other line separator via lineSeparator config property.
• version updates (spring-boot 2.7.8)
• spring boot 3 compatibility + tests
• added Opensearch 2.5 compatibility tests

• Optimization: Don't acquire lock if no scripts need to be executed (#172)
• Previously applied migration scripts are now checked for modifications and rejected if they've been modified after they were applied. The old behaviour can be restored by setting the new configuration parameter validateOnMigrate to false (default: true) (#155)
• version updates (spring-boot 2.7.7)
• added java 19 compatibility tests
• added spring boot 2.7 compatibility tests
• added Elasticsearch 8.6, 8.5, 8.4, 8.3, and 8,2 compatibility test
• added Opensearch 2.4, 2.3, 2.2, 2.1 and 2.0 compatibility tests
• It is now possible to set a baselineVersion to skip migrations with versions lower than the defined baselineVersion (#164)

• breaking change: drop org.elasticsearch.client.RestHighLevelClient and replace with org.elasticsearch.client.RestClient (LowLevelClient). This will drop the big transitive dependency org.elasticsearch:elasticsearch and opens compatibility to Elasticsearch 8 and OpenSearch.
• version updates (spring-boot 2.6.6)
• added spring boot 2.5 and 2.6 compatibility tests
• added java 17 and 18 compatibility tests
• added Elasticsearch 8.1, 8.0, 7.17, 7.16, 7.15, 7.14 and 7.13 compatibility tests
• added Opensearch 1.0, 1.1, 1.2 and 1.3 compatibility tests
• fixed issue #114

• fixed issue #36
• version updates (spring-boot 2.4.5)
• added java 16 compatibility tests
• added Elasticsearch 7.12 compatibility tests

• fixed issue #29
• fixed issue #27
• version updates (spring-boot 2.4.3)

• version upgrade elasticsearch-rest-high-level-client to 7.5.2 (Es version < 7.5.0 are no longer supported)
• remove Spring-Boot 1.5 and 2.0 support
• version updates (spring-boot 2.4.0)
• added spring-boot 2.4 compatibility tests

• version updates (spring-boot 2.3.5.RELEASE)

• version updates (spring-boot 2.3.0.RELEASE, elasticsearch 6.8.6, jackson 2.10.3, slf4j 1.7.30, reflections 0.9.12)
• added spring-boot 2.2 compatibility tests
• added spring-boot 2.3 compatibility tests

• new configuration parameter historyMaxQuerySize
• version updates (spring-boot 2.1.9.RELEASE; elasticsearch 6.8.3)
• ElasticsearchEvolutionInitializer logs now the migration duration

• version updates (spring-boot 2.1.3.RELEASE; elasticsearch 6.7.0)
• bugfix: support more than 10 migration scripts: now 1000.

• improved logging
• fixed classpath scanning in fat-jars like spring-boot

• initial version